[FieldTrip] simulated/sample data set for beamforming
alexander_nakhnikian at hms.harvard.edu
Sat Jun 11 19:38:32 CEST 2016
I also got some strange results from beamforming, although I was using
DICS, not LCMV. I'm working with EEG data so I don't know if we're
using the same recording setup. Jan-Mathijs suggested that there might
be problems with the forward model that are influencing the shape of the
beamformer. You can look at our exchange if it's still in your inbox.
Regarding modeling, I think fieldtrip has programs to simulated sensors
given a particular dipole distribution but I haven't used them. Another
approach would be to save the leadfields and apply them to a simulated
matrix of source-level oscillators. Set most of the "sources" to random
noise and place sinusoids with a known frequency and high SNR at only a
few locations (don't use identical oscillators for each source since
beamformers lose spatial resolution for strongly correlated sources). I
haven't tried this in fieldtrip, but I was able to do it with brainstorm
and the "dipoles" showed up at the correct locations when I applied an
inverse model to the simulated sensor data.
On 6/10/16 8:57 AM, Hassan Aleem wrote:
> Hi all,
> I have a pipeline for beamforming but I am getting unexpected results
> from my own data, for the sake of sanity I wanted to run some
> simulated data to see what I get. I am looking to do a
> timelockanalysis with the LCMV method, so can't use the tutorial data.
> Would appreciate any help!
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
Alexander Nakhnikian, Ph.D.
Clinical Research Training Program
VA Boston Healthcare System
Harvard Medical School
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the fieldtrip